视频目标跟踪技术综述  

Review of video target tracking technology

作　　者：李均利[1] 尹宽 储诚曦[2] 汪鸿年 LI Junli;YIN Kuan;CHU Chengxi;WANG Hongnian(Department of Computer Science,Sichuan Normal University,Chengdu,Sichuan 610101,China;Faculty of Electrical Engineering and Computer Science,Ningbo University,Ningbo,Zhejiang 315211,China)

机构地区：[1]四川师范大学计算机科学学院,四川成都610101 [2]宁波大学信息科学与工程学院,浙江宁波315211

出　　处：《燕山大学学报》2019年第3期251-262,共12页Journal of Yanshan University

基　　金：国家自然科学基金资助项目(61403266)

摘　　要：目标检测与跟踪是计算机视觉领域中的一个基本课题和研究热点,在军事领域、视频监控、人机交互、交通监测、精准制导等多个领域中有着重要的应用,具有重要的理论研究意义与实际应用价值。本文旨在对经典目标跟踪的技术流程以及技术发展现状作简要阐述。尤其对近些年研究热点,如相关滤波和深度学习的跟踪方法进行了概述,对其未来发展趋势进行了展望。另外,本文分析了多种方法的优缺点,提出了目标跟踪过程的主要挑战和难题,为相关研究人员提供参考和建议。Target detection and tracking is a basic subject and research hotspot in the field of computer vision. It has important applications in many fields such as military field, video surveillance, human-computer interaction, traffic monitoring, and precision guidance. It has important theoretical research significance and practical application value.The purpose of this article is to briefly explain the technical process of classical target tracking and the current state of technology development. In particular, the research hotspots in recent years, such as the methods of tracking correlation filtering and deep learning, are summarized and their future development trends are prospected. In addition, the article analyzes the advantages and disadvantages of various methods, the main difficulties and challenges in the target tracking process are proposed, besides, the reference and suggestions for relevant researchers are provided.

关 键 词：计算机视觉 目标跟踪 相关滤波 深度学习 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]